ATOM: Asynchronous Temporal Models
With Asynchronous Temporal Models (ATOM), Borealis AI researchers aim to build machine learning models capable of making inferences from asynchronous event sequences. Asynchronous event sequences are ubiquitous in personal banking applications – from various types of transaction data to client interactions with RBC banking services.
ATOM Research 🔎
This novel research agenda aims to support Borealis products in personal banking and risk management, as well as have scientific impact. This opens the door for our researchers to build technologies that leverage data from multiple channels, but it also syncs seamlessly with clients that use only a subset of these channels, in a manner that respects privacy and upholds principles of Responsible AI.
For example, Borealis AI’s technology currently powers an account cashflow feature in the RBC mobile app, and better asynchronous temporal models can help us make more accurate predictions, recognize when we are uncertain, and in the end deliver a better user experience.
Machine Learning for a better financial future 💫
The team’s commitment to creating real-world impact through scientific pursuit led to Borealis AI establishing a set of challenging North Star research problems, such as Asynchronous Temporal Models. The research team’s work is integral to the projects Borealis AI undertakes in the personal banking space and sits at the core of RBC’s overall innovation strategy.
Paper Spotlight: Meta Temporal Point Processes
One of the recently published papers, Meta Temporal Point Processes (ICLR 2023) focuses on a temporal point process (TPP), a stochastic process where its realization is a sequence of discrete events in time. In this paper, the researchers propose to train TPPs in a meta learning framework, where each sequence is treated as a different task, via a novel framing of TPPs as neural processes and introduce context sets to model TPPs as an instantiation of neural processes.
Motivated by attentive neural processes, the researchers also introduce local history matching to help learn more informative features, and demonstrate the potential of the proposed method on popular public benchmark datasets and tasks, and compare with state-of-the-art temporal point process methods.
Read the full Meta Temporal Point Processes (ICLR 2023) paper here.
Select ATOM Publications
-
Meta Temporal Point Processes
Meta Temporal Point Processes
W. Bae, M. O. Ahmed, F. Tung, and G. Oliveira. International Conference on Learning Representations (ICLR), 2023
Publication
-
Ranking Regularization for Critical Rare Classes: Minimizing False Positives at a High True Positive Rate
Ranking Regularization for Critical Rare Classes: Minimizing False Positives at a High True Positive Rate
*M. Kiarash, H. Zhao, M. Zhai, and F. Tung. The IEEE / CVF Computer Vision and Pattern Recognition Conference (CVPR), 2023
Publication
-
RankSim: Ranking Similarity Regularization for Deep Imbalanced Regression
RankSim: Ranking Similarity Regularization for Deep Imbalanced Regression
Y. Gong, G. Mori, and F. Tung. International Conference on Machine Learning (ICML), 2022
Publication
-
Self-Supervised Time Series Representation Learning with Temporal-Instance Similarity Distillation
Self-Supervised Time Series Representation Learning with Temporal-Instance Similarity Distillation
A. Hajimoradlou, L. Pishdad, F. Tung, and M. Karpusha. Workshop at International Conference on Machine Learning (ICML), 2022
Publication
-
Gumbel-Softmax Selective Networks
Gumbel-Softmax Selective Networks
M. Salem, M. O. Ahmed, F. Tung, and G. Oliveira. Workshop at Conference on Neural Information Processing Systems (NeurIPS), 2022
Publication
-
Training a Vision Transformer from scratch in less than 24 hours with 1 GPU
Training a Vision Transformer from scratch in less than 24 hours with 1 GPU
S. Irandoust, T. Durand, Y. Rakhmangulova, W. Zi, and H. Hajimirsadeghi. Workshop at Conference on Neural Information Processing Systems (NeurIPS), 2022
Publication
ATOM is integral to Borealis AI products and projects
NOMI Forecast: AI for digital money management
Incorporating AI into digital banking and empowering people to make better financial decisions is a compelling challenge. Borealis AI is creating tools like the newly released NOMI Forecast that help millions of RBC clients manage their finances.